Today's cloud computing systems offer users inexpensive and scalable access to a wide range of software services and resources that expedite development, deployment, and management of software and data. Different cloud vendors compete for these users via pricing, service capability and availability, scale, and ease of use, among other features. Despite their advantages, using any single cloud vendor alone limits user choice, results in vendor lock-in and "data gravity" (i.e., the storing of data with the vendor at which the user has previously stored data), and exposes users to greater risk of failures and privacy violations.
UCSB Computer Sciences Professors Chandra Krintz, Rich Wolski, Divy Agrawal, and Amr El Abbadi recently won a new NSF grant to address these limitations with new systems technologies that enable users to leverage multiple cloud infrastructures at once, safely and easily. In particular, it defines a new software abstraction for the scalable data management (datastore) layer, called DatGeo, that bridges geographically distributed cloud federations. The research will use DatGeo to develop new approaches for efficient transactions, partitioning and replication of data, and policy enforcement and mediation, across clouds.
As a result, DatGeo will shield user applications from the complexities associated with low-level federation of individual cloud services, while facilitating location and privacy control, increased reliability, and transparent cross-cloud use and portability. To enable wide spread use, the project will make its research artifacts and systems prototypes available as open source. In addition, the project will result in new course materials and activities that engage diverse students, new to computer science from local high-school and teaching-focused colleges, to introduce them to computer science as a potential career path.